INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

istries

-0.72

sic

-0.71

elist

-0.69

tips

-0.64

assies

-0.63

nesota

-0.63

gypt

-0.63

 opener

-0.63

 curfew

-0.62

matter

-0.61

POSITIVE LOGITS

itic

0.67

ITE

0.65

 privatization

0.64

Discussion

0.62

ux

0.60

ailing

0.60

ailed

0.60

urred

0.59

MON

0.59

cha

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.